A structural break is a concept in econometrics. A structural break appears when we see an unexpected shift in a (macroeconomic) time series. This can lead to huge forecasting errors and unreliability of the model in general[1]. This issue was popularised by David Hendry.
Contents |
In general, the CUSUM (cumulative sum) and CUSUM-sq (CUSUM squared) tests can be used to test the constancy of the coefficients in a model. The bounds test introduced by Hashem Pesaran, Yongcheol Shin, and Richard Smith can also be used.
For a linear model with one known single break in mean, the Chow test is often used. If the single break in mean is unknown, then the sup test may be appropriate. Other challenges are where there are:
The Chow test is not applicable for these situations[1].
For nonstationary process, there are many more challenges. For a cointegration model, the Gregory and Hansen test (1996) is used for one unknown structural break, and the Hatemi-J test (2007) is used for two unknown breaks.
There are several programs that can be used to find structural breaks, including R (open source) and Gauss.
If there are too many unknown breaks, then just assume the parameter to be time varying.
The latest method has been used by Bai and Perron (2003) in which multiple structural breaks can be automatically detected from data. The literature in this regard is very vast starting right from 1987 to 2010. Recently economists are going for both growth rate analysis and also econometric analysis in order to find break points one such way has been recommended by Chandan Mukherjee (2009).